{
  "schemaVersion": "1.0",
  "item": {
    "slug": "pinchbench",
    "name": "PinchBench",
    "source": "tencent",
    "type": "skill",
    "category": "开发工具",
    "sourceUrl": "https://clawhub.ai/olearycrew/pinchbench",
    "canonicalUrl": "https://clawhub.ai/olearycrew/pinchbench",
    "targetPlatform": "OpenClaw"
  },
  "install": {
    "downloadMode": "redirect",
    "downloadUrl": "/downloads/pinchbench",
    "sourceDownloadUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=pinchbench",
    "sourcePlatform": "tencent",
    "targetPlatform": "OpenClaw",
    "installMethod": "Manual import",
    "extraction": "Extract archive",
    "prerequisites": [
      "OpenClaw"
    ],
    "packageFormat": "ZIP package",
    "includedAssets": [
      "README.md",
      "SKILL.md",
      "assets/ai_blog.txt",
      "assets/quarterly_sales.csv",
      "crab.txt",
      "pyproject.toml"
    ],
    "primaryDoc": "SKILL.md",
    "quickSetup": [
      "Download the package from Yavira.",
      "Extract the archive and review SKILL.md first.",
      "Import or place the package into your OpenClaw setup."
    ],
    "agentAssist": {
      "summary": "Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually.",
      "steps": [
        "Download the package from Yavira.",
        "Extract it into a folder your agent can access.",
        "Paste one of the prompts below and point your agent at the extracted folder."
      ],
      "prompts": [
        {
          "label": "New install",
          "body": "I downloaded a skill package from Yavira. Read SKILL.md from the extracted folder and install it by following the included instructions. Then review README.md for any prerequisites, environment setup, or post-install checks. Tell me what you changed and call out any manual steps you could not complete."
        },
        {
          "label": "Upgrade existing",
          "body": "I downloaded an updated skill package from Yavira. Read SKILL.md from the extracted folder, compare it with my current installation, and upgrade it while preserving any custom configuration unless the package docs explicitly say otherwise. Then review README.md for any prerequisites, environment setup, or post-install checks. Summarize what changed and any follow-up checks I should run."
        }
      ]
    },
    "sourceHealth": {
      "source": "tencent",
      "status": "healthy",
      "reason": "direct_download_ok",
      "recommendedAction": "download",
      "checkedAt": "2026-04-30T16:55:25.780Z",
      "expiresAt": "2026-05-07T16:55:25.780Z",
      "httpStatus": 200,
      "finalUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=network",
      "contentType": "application/zip",
      "probeMethod": "head",
      "details": {
        "probeUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=network",
        "contentDisposition": "attachment; filename=\"network-1.0.0.zip\"",
        "redirectLocation": null,
        "bodySnippet": null
      },
      "scope": "source",
      "summary": "Source download looks usable.",
      "detail": "Yavira can redirect you to the upstream package for this source.",
      "primaryActionLabel": "Download for OpenClaw",
      "primaryActionHref": "/downloads/pinchbench"
    },
    "validation": {
      "installChecklist": [
        "Use the Yavira download entry.",
        "Review SKILL.md after the package is downloaded.",
        "Confirm the extracted package contains the expected setup assets."
      ],
      "postInstallChecks": [
        "Confirm the extracted package includes the expected docs or setup files.",
        "Validate the skill or prompts are available in your target agent workspace.",
        "Capture any manual follow-up steps the agent could not complete."
      ]
    },
    "downloadPageUrl": "https://openagent3.xyz/downloads/pinchbench",
    "agentPageUrl": "https://openagent3.xyz/skills/pinchbench/agent",
    "manifestUrl": "https://openagent3.xyz/skills/pinchbench/agent.json",
    "briefUrl": "https://openagent3.xyz/skills/pinchbench/agent.md"
  },
  "agentAssist": {
    "summary": "Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually.",
    "steps": [
      "Download the package from Yavira.",
      "Extract it into a folder your agent can access.",
      "Paste one of the prompts below and point your agent at the extracted folder."
    ],
    "prompts": [
      {
        "label": "New install",
        "body": "I downloaded a skill package from Yavira. Read SKILL.md from the extracted folder and install it by following the included instructions. Then review README.md for any prerequisites, environment setup, or post-install checks. Tell me what you changed and call out any manual steps you could not complete."
      },
      {
        "label": "Upgrade existing",
        "body": "I downloaded an updated skill package from Yavira. Read SKILL.md from the extracted folder, compare it with my current installation, and upgrade it while preserving any custom configuration unless the package docs explicitly say otherwise. Then review README.md for any prerequisites, environment setup, or post-install checks. Summarize what changed and any follow-up checks I should run."
      }
    ]
  },
  "documentation": {
    "source": "clawhub",
    "primaryDoc": "SKILL.md",
    "sections": [
      {
        "title": "PinchBench Benchmark Skill",
        "body": "PinchBench measures how well LLM models perform as the brain of an OpenClaw agent. Results are collected on a public leaderboard at pinchbench.com."
      },
      {
        "title": "Prerequisites",
        "body": "Python 3.10+\nuv package manager\nOpenClaw instance (this agent)"
      },
      {
        "title": "Quick Start",
        "body": "cd <skill_directory>\n\n# Run benchmark with a specific model\nuv run benchmark.py --model anthropic/claude-sonnet-4\n\n# Run only automated tasks (faster)\nuv run benchmark.py --model anthropic/claude-sonnet-4 --suite automated-only\n\n# Run specific tasks\nuv run benchmark.py --model anthropic/claude-sonnet-4 --suite task_01_calendar,task_02_stock\n\n# Skip uploading results\nuv run benchmark.py --model anthropic/claude-sonnet-4 --no-upload"
      },
      {
        "title": "Available Tasks (23)",
        "body": "TaskCategoryDescriptiontask_00_sanityBasicVerify agent workstask_01_calendarProductivityCalendar event creationtask_02_stockResearchStock price lookuptask_03_blogWritingBlog post creationtask_04_weatherCodingWeather scripttask_05_summaryAnalysisDocument summarizationtask_06_eventsResearchConference researchtask_07_emailWritingEmail draftingtask_08_memoryMemoryContext retrievaltask_09_filesFilesFile structure creationtask_10_workflowIntegrationMulti-step API workflowtask_11_clawdhubSkillsClawHub interactiontask_12_skill_searchSkillsSkill discoverytask_13_image_genCreativeImage generationtask_14_humanizerWritingText humanizationtask_15_daily_summaryProductivityDaily digesttask_16_email_triageEmailInbox triagetask_17_email_searchEmailEmail searchtask_18_market_researchResearchMarket analysistask_19_spreadsheet_summaryAnalysisSpreadsheet analysistask_20_eli5_pdf_summaryAnalysisPDF simplificationtask_21_openclaw_comprehensionKnowledgeOpenClaw docs comprehensiontask_22_second_brainMemoryKnowledge management"
      },
      {
        "title": "Command Line Options",
        "body": "OptionDescription--modelModel identifier (e.g., anthropic/claude-sonnet-4)--suiteall, automated-only, or comma-separated task IDs--output-dirResults directory (default: results/)--timeout-multiplierScale task timeouts for slower models--runsNumber of runs per task for averaging--no-uploadSkip uploading to leaderboard--registerRequest new API token for submissions--upload FILEUpload previous results JSON"
      },
      {
        "title": "Token Registration",
        "body": "To submit results to the leaderboard:\n\n# Register for an API token (one-time)\nuv run benchmark.py --register\n\n# Run benchmark (auto-uploads with token)\nuv run benchmark.py --model anthropic/claude-sonnet-4"
      },
      {
        "title": "Results",
        "body": "Results are saved as JSON in the output directory:\n\n# View task scores\njq '.tasks[] | {task_id, score: .grading.mean}' results/0001_anthropic-claude-sonnet-4.json\n\n# Show failed tasks\njq '.tasks[] | select(.grading.mean < 0.5)' results/*.json\n\n# Calculate overall score\njq '{average: ([.tasks[].grading.mean] | add / length)}' results/*.json"
      },
      {
        "title": "Adding Custom Tasks",
        "body": "Create a markdown file in tasks/ following TASK_TEMPLATE.md. Each task needs:\n\nYAML frontmatter (id, name, category, grading_type, timeout)\nPrompt section\nExpected behavior\nGrading criteria\nAutomated checks (Python grading function)"
      },
      {
        "title": "Leaderboard",
        "body": "View results at pinchbench.com. The leaderboard shows:\n\nModel rankings by overall score\nPer-task breakdowns\nHistorical performance trends"
      }
    ],
    "body": "PinchBench Benchmark Skill\n\nPinchBench measures how well LLM models perform as the brain of an OpenClaw agent. Results are collected on a public leaderboard at pinchbench.com.\n\nPrerequisites\nPython 3.10+\nuv package manager\nOpenClaw instance (this agent)\nQuick Start\ncd <skill_directory>\n\n# Run benchmark with a specific model\nuv run benchmark.py --model anthropic/claude-sonnet-4\n\n# Run only automated tasks (faster)\nuv run benchmark.py --model anthropic/claude-sonnet-4 --suite automated-only\n\n# Run specific tasks\nuv run benchmark.py --model anthropic/claude-sonnet-4 --suite task_01_calendar,task_02_stock\n\n# Skip uploading results\nuv run benchmark.py --model anthropic/claude-sonnet-4 --no-upload\n\nAvailable Tasks (23)\nTask\tCategory\tDescription\ntask_00_sanity\tBasic\tVerify agent works\ntask_01_calendar\tProductivity\tCalendar event creation\ntask_02_stock\tResearch\tStock price lookup\ntask_03_blog\tWriting\tBlog post creation\ntask_04_weather\tCoding\tWeather script\ntask_05_summary\tAnalysis\tDocument summarization\ntask_06_events\tResearch\tConference research\ntask_07_email\tWriting\tEmail drafting\ntask_08_memory\tMemory\tContext retrieval\ntask_09_files\tFiles\tFile structure creation\ntask_10_workflow\tIntegration\tMulti-step API workflow\ntask_11_clawdhub\tSkills\tClawHub interaction\ntask_12_skill_search\tSkills\tSkill discovery\ntask_13_image_gen\tCreative\tImage generation\ntask_14_humanizer\tWriting\tText humanization\ntask_15_daily_summary\tProductivity\tDaily digest\ntask_16_email_triage\tEmail\tInbox triage\ntask_17_email_search\tEmail\tEmail search\ntask_18_market_research\tResearch\tMarket analysis\ntask_19_spreadsheet_summary\tAnalysis\tSpreadsheet analysis\ntask_20_eli5_pdf_summary\tAnalysis\tPDF simplification\ntask_21_openclaw_comprehension\tKnowledge\tOpenClaw docs comprehension\ntask_22_second_brain\tMemory\tKnowledge management\nCommand Line Options\nOption\tDescription\n--model\tModel identifier (e.g., anthropic/claude-sonnet-4)\n--suite\tall, automated-only, or comma-separated task IDs\n--output-dir\tResults directory (default: results/)\n--timeout-multiplier\tScale task timeouts for slower models\n--runs\tNumber of runs per task for averaging\n--no-upload\tSkip uploading to leaderboard\n--register\tRequest new API token for submissions\n--upload FILE\tUpload previous results JSON\nToken Registration\n\nTo submit results to the leaderboard:\n\n# Register for an API token (one-time)\nuv run benchmark.py --register\n\n# Run benchmark (auto-uploads with token)\nuv run benchmark.py --model anthropic/claude-sonnet-4\n\nResults\n\nResults are saved as JSON in the output directory:\n\n# View task scores\njq '.tasks[] | {task_id, score: .grading.mean}' results/0001_anthropic-claude-sonnet-4.json\n\n# Show failed tasks\njq '.tasks[] | select(.grading.mean < 0.5)' results/*.json\n\n# Calculate overall score\njq '{average: ([.tasks[].grading.mean] | add / length)}' results/*.json\n\nAdding Custom Tasks\n\nCreate a markdown file in tasks/ following TASK_TEMPLATE.md. Each task needs:\n\nYAML frontmatter (id, name, category, grading_type, timeout)\nPrompt section\nExpected behavior\nGrading criteria\nAutomated checks (Python grading function)\nLeaderboard\n\nView results at pinchbench.com. The leaderboard shows:\n\nModel rankings by overall score\nPer-task breakdowns\nHistorical performance trends"
  },
  "trust": {
    "sourceLabel": "tencent",
    "provenanceUrl": "https://clawhub.ai/olearycrew/pinchbench",
    "publisherUrl": "https://clawhub.ai/olearycrew/pinchbench",
    "owner": "olearycrew",
    "version": "1.0.0",
    "license": null,
    "verificationStatus": "Indexed source record"
  },
  "links": {
    "detailUrl": "https://openagent3.xyz/skills/pinchbench",
    "downloadUrl": "https://openagent3.xyz/downloads/pinchbench",
    "agentUrl": "https://openagent3.xyz/skills/pinchbench/agent",
    "manifestUrl": "https://openagent3.xyz/skills/pinchbench/agent.json",
    "briefUrl": "https://openagent3.xyz/skills/pinchbench/agent.md"
  }
}